General ideas and concepts in this article
Understanding how and why a prospect moves for the consensus is a complex topic, but concepts from studying social systems and networks can help us understand it.
Categorizing information helps us to understand what others know and how to use it when trying to make accurate predictions.
Getting outside of our own viewpoint is a key to understanding complex systems like the draft and trying to form a broader view of possible outcomes.
Information cascades can create dangerous feedback loops that create market inefficiency and lead us away from “true value”.
We are social creatures with a keen awareness for what others think and do, which makes independent judgment and forming predictions tricky.
Differing goals and incentives can complicate what people share or use in mock drafts/big boards.
The quality of markets is determined by the quality and depth of information within them, which means we can identify where markets are potentially inefficient.
Depending on your goal, it’s important to understand how to use these concepts when trying to make the best prediction possible.
Information of Many Stripes
The Rise of Chandler Zavala
On March 21st, The Athletic’s draft analyst Dane Brugler tweeted about North Carolina State OG Chandler Zavala as a player with day two tape who he believes will be the highest non-Combine invitee drafted. It was a surprising and seemingly influential tweet that could serve as a catalyst for Zavala in how he is viewed by the consensus. But would a move upwards be rational or accurately reflect what the NFL thinks of Zavala?
Prior to Brugler’s tweet there were six instances of Zavala appearing in mock drafts or big boards according to nflmockdraftdatabase.com. He was ranked 236th overall in a 2/13 board, 289th overall in a 2/22 board, and appeared in four team focused mock drafts as a 5th round pick (2x) and a 7th round pick (2x).
In the past ten days, he has appeared in 10 more team mock drafts along with drawing plenty of tweets from draftniks about his abilities. As it stands, Zavala is still 194th overall on the consensus board, but if the current trend is any indication he will continue to go upwards as more boards get updated. But this is a player who was basically unranked by the consensus prior to February and did not receive the interest from teams to draw either a Senior Bowl or Combine invitation. What happened?
Categorizing Information - Private and Public
Let’s start with defining terms about what specific types of information mean. We can break up the information someone has when making a decision between two options - this information is often described as “signals” about the state of the world or a particular thing.
Private Information
This information can be viewed as what someone would have if they were operating in complete isolation and independence from all other sources. Suppose I asked you to evaluate USC defensive end Jim Jackson. You have no knowledge of his ranking, you’ve seen nobody talk about him, and there is generally nothing you can draw on from others. You could use his stats, his size, his athleticism, his tape, etc to judge him but your formulation of a judgment based on these things would be private to you. Nobody would know at that point if you think he’s a first rounder or an undrafted free agent.
Public Information
This information is everything else out there that could be used to assist in forming a judgment on Jim Jackson. If I message you and say, “Watch this Jim Jackson guy! He’s a first rounder to me!” or you see a bunch of mock drafts putting Jim Jackson high, then applying any of that would be utilizing public information.
Most things in life are not decided solely by private or public information of course. We are constantly tuning our decisions to account for both types of information. Much of this decision-making process is not in our conscious mind when we are reasoning about what types of choices we should make. The environment we are operating within, our personal goals, and general variance within our decision-making processes can all play a role in how we divide our private and public information.
If I provide you with Chandler Zavala as a player to come up with an estimate on? There will be an incredible number of variables to input, run through your mental black box, and then output on the other side as a singular estimate. This makes trying to apply some of these more simplistic models difficult in real world situations, but let’s assume we can compartmentalize here and eliminate some of these more “human” variables.
Information Cascades: Failure to Deploy Private Information
This concept of information cascades is something that can be applied broadly and should be considered as a possibility for plenty of prospects, but consider this in the context of how we are interpreting the information we’ve received about Zavala so far.
What are information cascades?
To provide a specific example that has been used fairly regularly to explain information cascades and the way private and public information function in the wild, I’ll use this quote from Networks, Crowds, and Markets: Reasoning about a Highly Connected World.
“As a first example, suppose that you are choosing a restaurant in an unfamiliar town, and based on your own research about restaurants you intend to go to restaurant A. However, when you arrive you see that no one is eating in restaurant A while restaurant B next door is nearly full. If you believe that other diners have tastes similar to yours, and that they too have some information about where to eat, it may be rational to join the crowd at B rather than follow your own information.
To see how this is possible, suppose that each diner has obtained independent but imperfect information about which of the two restaurants is better. Then if there are already many diners at restaurant B, the information that you can infer from their choices may be more powerful than your own private information, in which case it would in fact make sense for you to join them regardless of your own private information. In this case, we say that herding or an information cascade, has occurred.”
In this hypothetical world, we could know that restaurant A and restaurant B are of equal quality, but if the first person to arrive and make the decision were to choose restaurant B, then it could trigger a cascade of follow-up decisions by everyone who arrives afterwards and has to make the same decision.
We could also see the exact opposite play out where everyone goes to restaurant A. This process would happen not as a function of the true quality of either restaurant, but rather the path dependence (what happened in the past) of the prior decision-makers. Of course, the limitation to this model is that it assumes consistency of choice (we are not this rational), a certain degree of reliance on private and public information, and a particular goal (eating at the best restaurant). Our decision-making process when studying players, ranking them, or sharing our opinions on them publicly is not this simple.
Before we can apply the concepts directly to Zavala’s situation, let’s look at two more things to really nail down information cascades. The first is a video from St. Olaf professor of economics Ashley Hodgson explaining how an information cascade can take place using an example of individuals pulling marbles out of a bag. The second is a more technical definition of the ingredients to create an information cascade that comes from the aforementioned book about networks.
The Marble Example
Information Cascade Ingredients
(a) There is a decision to be made - for example, whether to adopt a new technology, wear a new style of clothing, eat in a new restaurant, or support a particular political position.
(b) People make the decision sequentially, and each person can observe the choices made by those who acted earlier.
(c) Each person has some private information that helps guide their decision.
(d) A person can’t directly observe the private information that other people know, but he or she can make inferences about this private information from what they do.
Private Information and Uncertainty
Understanding why we would make a decision about the bag of marbles or restaurant we choose is difficult without trying to understand those aspects of private and public information, where they matter, and how much we have.
Suppose I were to drop you off in front of those two restaurants and you have no private information. All you can see is that one restaurant is bustling and the other is basically empty. The rational choice here would be to go to the busy restaurant because you have no other information to apply.
Suppose I were to drop you off again, but this time you’ve eaten at restaurant A for years and know privately that you will always prefer it to restaurant B. Your private information would outweigh the public information and you would unfailingly select restaurant A.
The role your private information plays in your decision-making process is tied to how much confidence you have about the choice you are making. When you have no information or very little it will make sense to follow the crowd, but when you have a ton of private information that is a trustworthy indicator then you would be better off utilizing that regardless of what you see/hear from others.
This aspect of when to use private information and when to adjust away from it is difficult to fully apply because of our inherent biases, the role of social influence and social psychology, and the general difficulty of calibrating in an inherently messy world full of incomplete information.
Differing Goals and Their Influence
Another limitation within those models of a bag of marbles or two competing restaurants is the aspect of different end goals for those participating. In reality, the diners aren’t going to all judge their options based on the same dimensions. One may be searching for a restaurant with a certain atmosphere, another could be searching for the highest quality food, and another could be looking at what provides the best value relative to the price.
Consider this with the placement of Zavala in mock drafts, rankings, and big boards. One draftnik may be grading him purely based upon what he thinks Zavala will do in his rookie year rather than trying to project three years down the line. Another may be trying to project where Zavala is most likely to be selected rather than what his likely NFL outcome will be once he gets on a roster. Another may consider positional value when trying to slot him onto a board rather than just going by “raw talent relative to peers”.
These varying goals add even more complexity to public information that is available, as you can’t know the exact rationale behind the opinions of all individual peoples that compose the consensus. There is also an element of timing, who shares what with who, and status/trust involved in the way we receive and interpret signals from public information.
Categorizing Information Part 2 - Value and Timing
Categorizing information comes in many forms, and we would do well to remember the full spectrum of information available when we consider what is “public”. We can use certain context clues to help develop a picture of what information other parties have and how they have used it, and doing this can help us balance the most recent information with what was presumably known in the past.
To categorize Zavala, we can choose among many different reference classes (groups in which Zavala is a member). These reference classes all tell us something a little bit different about who he is and what we can do to formulate an accurate judgment. We can look at his size, his athletic testing data, his career trajectory throughout college, his statistical output, the all-star game he attended, and many other similarly valuable groups to pull from. Here are the ones that really catch my attention.
Zavala does not appear to have been invited to the Senior Bowl.
When you consider the makeup of the Senior Bowl roster, it’s worth considering the timeline in which the roster is composed, the information that is taken in when offering invitations, and the relative depth of the class at each position (how many roster spots are available). It is also worth considering the historical success of the roster relative to the rosters of other all-star games. This information is incredibly telling because we can make some assumptions about what others know, even if we don’t have access to their private information.
These are my assumptions when considering what I know about the process of Senior Bowl invitations. Bruce Feldman wrote an insightful article about their process for selecting players that composes much of this knowledge, but some of this has been accumulated from just watching the way they build a roster and their past results.
The Senior Bowl is looking to invite the best available players at every position within the constraints of the roster and player eligibility, but there is some consideration taken to give opportunities to players who are at a lower level of competition without “best on best” exposures throughout the year.
Class to class variation in positional depth will lead to some positions being weaker or stronger, but generally speaking players who are viewed as top four round players are safe bets to be invited.
The invitation process happens in early November, which means we are missing key information on prospect testing data, all-star performance, medical, and character outside of what is known in the fall.
The Senior Bowl staff has calls with a large number of teams across the league to crosscheck their own work, and these calls give them an incredible amount of insight into how teams view the draft class in the Fall.
The East-West Shrine Bowl managed to poach two players from the Senior Bowl this past season, with one of them being Zay Flowers and the other one being unknown (at least to me). This means that Zavala *could* have been invited and passed up a Senior Bowl invite, but the odds of this do not seem high.
This roster is one of the best indicators we have for how teams collectively view these players entering the pre-draft process and there is a significant amount of historical data that backs this up as far as number of players drafted.
If Zavala was not invited to the Senior Bowl, we can assume that he was not viewed as one of the best players at his position by most teams in the league entering early November, assuming the sample (team calls) was not an unlucky or biased draw.
Zavala was invited to the East-West Shrine Bowl.
Let’s not get too pessimistic! The East-West Shrine Bowl has the second best track record as far as players being drafted or signing NFL contracts, and most of the players who attend the game will at least make camp. The assumptions we can make about the Senior Bowl are mostly the same assumptions we can make about the East-West Shrine Bowl, although I will admit that I am much less familiar with Shrine Bowl strategy and invitation processes than the Senior Bowl.
It’s impossible to know if Zavala was valued as a top four round type of option by the Shrine Bowl staff (barring them telling us) because they didn’t get to make the first picks in selecting their roster. Here are a couple of assumptions that I would make about the Shrine Bowl based on their past success.
Any player invited to the Shrine Bowl has a decent shot at being drafted, and at worst they should be a quality undrafted free agent. An occasional player slips below this, but generally speaking there is a high level of talent at the event.
If there is a particularly deep position group that has top four or five round players that can’t squeeze onto the Senior Bowl roster due to space limitations, the Shrine Bowl is very likely to scoop them up.
There are only a few players that attend this event who make their way into the top 100 picks, but it can happen. A player emerging as a possible top 100 pick from a smaller all-star event like the College Gridiron Showcase or Hula Bowl would be surprising, but it is a fairly common occurrence that someone makes this push from the Shrine Bowl each year.
For context, here are the numbers from 2022 for Senior Bowl and Shrine Bowl players. This reference class (which all-star game a player attended) gives us significant insight. Note: Specialists are not included in these numbers.
Senior Bowl
There were 127 players in total. 107 were drafted. 18 were undrafted free agents. 2 only received tryout opportunities entering May. Of the players selected in the top 100, 43 of them came from the Senior Bowl.
Shrine Bowl
There were 128 players in total. 45 were drafted. 74 were undrafted free agents. 6 received tryout opportunities. 3 received no NFL contract. Of the players selected in the top 100, 2 of them came from the Shrine Bowl.
Comparing Head to Head
Drafted
Senior Bowl - 84.26% (107/127)
Shrine Bowl - 35.16% (45/128)
Undrafted Free Agent
Senior Bowl - 14.17% (18/127)
Shrine Bowl - 57.81% (75/128)
Drafted within the first 100 picks
Senior Bowl - 33.86% (43/127)
Shrine Bowl - 1.56% (2/128)
Okay - so let’s assume we know nothing else about Chandler Zavala. We don’t know his position, we don’t know his level of competition, we don’t know his testing, his tape grade, or anything else. If we were to make a guess based purely on the information provided in an all-star game reference class (which reflects largely what teams think in November), then we should assume that the odds of Zavala being a top 100 draft pick are fairly low. Let’s continue on this for a moment.
Zavala was not invited to the NFL Combine.
For the same reasons that we would have concerns about Zavala appearing to not be invited to the Senior Bowl, we can point towards some concerns as a result of him not being invited to the NFL Combine. The Combine’s invitation process, to my knowledge, occurs slightly later than the all-star game invitations, and it happens more so in December and January.
This invitation process is done by National Football Scouting (NFS), who puts on the NFL Combine, but it uses a large pool of information across the league to decide who to invite and who to pass on. There may be some instances of gamesmanship involved where teams may try to sneak a player under the radar, but for the most part this process is incredibly effective and will catch most players who are likely to be drafted as of the invitation date.
I won’t go through the numbers again like I did with the Senior Bowl and Shrine Bowl, but it is worth noting that being a combine snub is a very notable ding on a player’s odds of being drafted. The vast majority of players who are drafted come from the combine group and Zavala failing to receive that invite is another sign of a potential low consensus ranking by the league entering the pre-draft process.
Zavala did not appear to draw much attention until very recently.
This one gets very interesting, because we have to take the timing aspect into account along with trying to figure out why he has had a sudden surge. The consensus can often reflect a lot of good information and provide much insight for us, but as we have fewer samples to draw from (fewer draftniks watching or ranking Zavala) the consensus starts to rapidly lose its value.
We have to pose this question. Why is Zavala coming on so late, but the early signals like the Senior Bowl and Combine missed him?
Is it because he was undervalued by scouts but well received by coaches? Is it him having an outstanding pre-draft process? Is he interviewing tremendously well? Is the recent signal (Brugler’s tweet and ranking) an accurate reflection of Zavala’s status among NFL decision-makers? Are we looking at a scenario where this idea of an information cascade could be impacting us?
What does Brugler know?
A lot of our assumptions about the validity of Zavala’s late rise hinge on the value of Brugler’s information, and the information of anyone else who began to clamor about Zavala before or after Brugler tweeted that. We have to understand what role he plays in moving the market, so to speak. We have to ask ourselves questions like these ones.
Why did it take until late March for Brugler to rank Zavala so highly?
If Zavala has been gaining traction as a prospect for teams, is this part of what put Brugler onto him and led to such a large move up his rankings?
How much of Brugler’s process is driven by private information and how much is driven by public information? Are league sources influencing his ranking of Zavala?
What goal do we think he has? Is he focused on trying to mirror where players will be selected? What is his version of “having a good board” post-draft or in the future?
This is a peculiar time of year because some draftniks are contacted by agents who ask them to spread certain pieces of information about players. The draftnik gets to sound like an insider because they can tweet the notorious “league sources tell me that such and such…” and agents get positive PR for their players. I don’t believe Brugler is doing this, but it’s just one example of where information can fly around that isn’t independent or rooted in true “accuracy seeking” behavior.
Here are some assumptions I would make about Brugler’s goals when formulating and sharing opinions. I don’t know him personally, I don’t know his process in a detailed manner, and I mean no offense by stating these assumptions. If you are more informed about his process or hold different beliefs about how he interprets and uses information, feel free to substitute them here and come to different conclusions about the validity of Zavala’s rise.
As a member of the media who is providing information to a large audience, and as someone who makes his living providing it, he has incentives to closely match what the NFL as a whole believes about a player. Being a contrarian or way off consensus has the potential to do reputational damage.
If he lands with an extremely low or high grade on a player, he has an incentive to reel it in or do a significant amount of due diligence to verify it.
This combination of incentives makes me believe that he has heard from trustworthy sources that Zavala will be drafted relatively early, despite the conflicting signals of no Combine or Senior Bowl invite.
I’m going to assume that, while Brugler obviously verified the grade through watching plenty of tape and doing his own due diligence on the player, he is also hearing things and factoring it into his projection. This assumption could be incorrect. He could be going out on a limb here. But assuming these assumptions are correct, it leads us to some other interesting questions.
What things would cause Zavala to rise after the all-star game and Combine invites went out?
What role do experts like Brugler play in the draftnik ecosystem? Can they function as catalysts for movement up or down boards?
What happens if Brugler stays silent about Zavala?
What social mechanisms are at play once Brugler kicks things off?
Markets, Networks, Judgment, Sociality, and Consensus
Experts and Catalysts
Networks and markets are usually incredibly complex and can operate in weird ways because of path dependency, so what creates a significant “stock up” or “stock down” moment in one run of the universe could have played out much differently if we were to turn back the clock and run it again with very minor changes in early initial conditions. Here are some of the ways where we never get “Zavala’s jump” and I am likely writing this article about a different player, or perhaps not writing it at all.
Zavala’s petition for a sixth season is denied by the NCAA. He is in the 2022 NFL Draft instead.
He suffers a severe injury early in the season and either returns for a seventh season or declares for the draft with no recent film or participation in the pre-draft process.
Dane Brugler does not tweet about him on March 21st and he stays relatively unnoticed by the community of draftniks.
He earns an invite to the NFL Combine and performs admirably, spurring attention earlier in the process.
Daniel Jeremiah or Mel Kiper calls him a potential top 100 pick in December after they get tipped off by a league source about his film. He quickly vaults into the sightlines of draftniks as a result.
I’ll be forward here. I’m not knowledgeable enough on network theory to tell you whether or not the consensus certainly moves because an “in the know” expert like Brugler or Jeremiah tweets. I’ve read and listened to conflicting things about what propels certain ideas/beliefs in networks while others wither out and die immediately. Oftentimes it’s just random and unpredictable in advance. My personal belief is that, depending on the player and the current level of information available on them, a strong endorsement from those perceived in the social hierarchy as trustworthy or “in the know” can push a player up significantly.
This is anecdotal, of course, and it is solely based on my time watching draft coverage and seeing the ways players move up or down. It would take a more extensive and rigorous analysis to say these market movers are more fact than narrative. Nobody remembers the times where Brugler or Jeremiah endorses a player to an audience of crickets.
In the scenario where these individuals are market movers, we should sometimes see this reflected in the movement of players on consensus boards. This does not necessarily mean that the individual has perfectly accurate beliefs about the player, but it does mean that the player can be exposed to a larger audience than they were previously exposed to, which can in turn lead to more positive feedback, repeat repeat.
One example that is interesting is Jeremiah’s high initial ranking of Georgia Tech defensive end Keion White. White debuted at 8th overall on Jeremiah’s Big Board 1.0 on January 31st. This chart from Grinding The Mocks is why I believe that market movers do exist. Look at the turn of January into February.
Here is another example of a potential catalyst event. Texas Tech’s Tyree Wilson skyrocketed in June after NFS ranked him as the top senior in the class. Prior to the ranking, Wilson had been virtually unknown to the draftnik community, and he did not emerge until well after teams received those early grades from NFS. He appeared in no mock drafts before this despite apparently having a 2021 performance worth being ranked as the top senior.
This isn’t a dig at those who were adjusting for the emergence of White or Wilson based on those catalyst events (NFS grades, Jeremiah’s big board). Most people who cover the draft online or produce player rankings/mock drafts are not digging around for talent at less acclaimed power five programs. A huge junior Texas Tech defender might not catch their eye when they are watching games like LSU vs. Alabama or USC vs. Notre Dame. When they begin their “too early” mock drafts they’re more likely to play a game of shuffling from what others have done rather than go explore the range of options.
This is a rational and optimal decision based on their time, resources, ability, and goals. But it does open us up to those information cascade problems that were mentioned earlier in the article. So let’s ask this question - what happens when an event or catalyst does not happen?
The Counterfactual: Brugler Stays Silent
Understanding the impact that something has on the system as a whole is sometimes not measurable in a way that doesn’t resort to hypotheticals and “counterfactuals” about what twisting turns of history would have happened otherwise. But let’s entertain the idea for a moment.
What would have happened to Zavala if Brugler had not tweeted about him or ranked him in his top 100?
We could assume that, if his assessment is accurate and the NFL feels strongly positive about Zavala, it will be reflected later on in the process by someone like Jeremiah, Kiper, McShay, or another expert that tends to have a good read on the relative rankings of players by draft night. But we have no way of knowing when this would take place. It likely would not have happened on March 22nd or one of the days shortly after it. Based on the lack of other major updates from similarly established experts like Brugler, we likely would have continued to sit in our prior state of information.
Zavala was ranked 236th when I began writing this article and moved up after the update adjusting for Brugler’s board, and prior to Brugler tweeting about him he was ranked around 308th overall. In this “no tweet” world, Zavala is still sitting around 308th overall.
These counterfactuals create some dilemmas when we consider what the consensus reflects relative to time and start conditions though. If Zavala were to appear in the top 100 for multiple analysts the night before the draft, the collective consensus wouldn’t have as much time to adjust and adapt to the new information, which would create an inefficiency between the “true value” of the player as defined by the pool of information known to the NFL and the “consensus value” of the player that reflects all publicly known information aggregated online from draftniks.
Alternatively, suppose that Zavala had been ranked in the top 100 by Brugler heading into the all-star circuit. What would his path have looked like if we dropped him into the “public eye” much earlier and he was seen and ranked by a larger number of draftniks? Would he have stayed ranked in the top 100 throughout the process without being a Combine invite? Would he have taken a moonshot up the board? Would he have gradually slid down?
Social Proof: Looking Where Others Look
It’s sometimes easy to bash on the consensus or “groupthink” because of the areas where it leads us astray, but often we look to the opinions and information of others with good reason. We’re naturally designed to. If we were socially inept, stumbling and bumbling along without a care for what our fellow humans think, we would face severe repercussions. Those repercussions would come in a variety of forms, from social ostracization to harm to life and limb.
In The Secrets to Our Success, Harvard’s Joseph Henrich discusses the ways in which our culture has allowed us to transmit ideas and optimize for survival through our shared intelligence. Most of the book is largely outside of the scope of this article, but this quote in regards to success or failure depending on our understanding of social and cultural practices is worth noting when we consider what we gain from information sharing. Bolded for emphasis.
“We saw big brained explorers repeatedly flounder in environments ranging from the Arctic to the Australian outback. As our heroes sought to confront the recurrent challenges faced by our paleolithic ancestors, like finding food and water, they struggled. No foraging modules fired up and no fire-making instincts kicked in. Mostly, they just fell ill and died as a result of blunders that any local, indigenous adolescent equipped with cultural know-how inherited from earlier generations could easily have avoided. It’s not merely that people in modern society need culture to survive. Hunter-gatherers, as well as other small-scale societies studied by anthropologists, are massively dependent on large scale bodies of cultural know-how, relating to tracking, food processing, hunting and tool manufacture. This expertise is often complex, well-adapted to local challenges, and not casually well understood by most practitioners… All human societies, whether they live as hunter-gatherers or not, are entirely dependent on culture.”
How does this apply to our decision-making processes? In harsher environments where much steeper penalties were paid for poor decisions, you would die if you failed to heed the information that exists in the minds of others. If you eat the unknown purple fruit that everyone else is avoiding, you are more likely to die than those who avoid it. If everyone starts running at the same time, it would behoove you to do the same lest you end up being a lion’s afternoon snack. Passing on the purple fruit is a small price to pay for avoiding being deathly ill, just like running away when there may be nothing is a small price to pay to avoid death by lion.
Over the extremely long course of our development as a species, we’ve become hardwired to act in certain ways. Oftentimes, we don’t even realize how deeply ingrained these norms and impulses are to us until they are poked and prodded.
James Surowiecki’s The Wisdom of Crowds did an excellent job covering the idea of social proof and how it was tested by early social psychologists, and much of these same concepts cross over from avoiding blunders in prehistory to how we react in social situations to this day. Here’s the passage on it.
“In 1968, the social psychologists Stanley Milgram, Leonard Bickman, and Lawrence Berkowitz decided to cause a little trouble. First they put a single person on a street corner and had him look up at an empty sky for sixty seconds. A tiny fraction of the passing pedestrians stopped to see what the guy was looking at, but most just walked past. Next time around, the psychologists put five skyward-looking men on the corner. This time, four times as many people stopped to gaze at the empty sky. When the psychologists put fifteen men on the corner, 45 percent of all passers by stopped, and increasing the cohort of observers yet again made more than 80 percent of pedestrians tilt their heads and look up.
This study appears at first glance, to be another demonstration of people's willingness to conform. But in fact it illustrated something different, namely the idea of "social proof", which is the tendency to assume that if lots of people are doing something or believe something, there must be a good reason why.
This is different from conformity: people are not looking up at the sky because of peer pressure or a fear of being reprimanded. They're looking up at the sky because they assume - quite reasonably - that lots of people wouldn't be gazing upward if there weren't something to see. That's why the crowd becomes more influential as it becomes bigger: every additional person is proof that something important is happening. And the governing assumption seems to be that when things are uncertain, the best thing to do is just to follow along. This is actually not an unreasonable assumption. After all, if the group usually knows best (as I've argued it often does), then following the group is a sensible strategy. The catch is that if too many people adopt that strategy, it stops being sensible and the group stops being smart.”
Does this sound a little familiar? It’s similar to the situation with the information cascade of marbles or restaurants where we have our private information (nothing is happening) but there is sufficient public information (people looking upwards) to warrant us to join in. This happens on a much more intuitive and instinctive level than choosing a restaurant or guessing the most likely marble color in a bag, but the function of private and public information is similar.
If you were driving down the road and the car in front of you slowed down while moving slightly to the left, what would you do? Would you continue at your current speed without consideration for their action?
Personally, in all the times I’ve noticed this and reacted to it, I’ve slowed down and followed their lead. So far I’ve never been wrong in assuming that their information tells me something, usually that a person or thing to be avoided is on the right side of the road. I once avoided hitting a flooded section of a road at full speed because someone was in front of me and slowed down. It would have been irrational of me to not use their information and without that information I probably would have wrecked.
Put into the context of a Tyree Wilson or Keion White, if everyone is paying attention to them, you should probably do the same if your incentives are aligned. Social proof suggests they are more worthy of your time than an unknown prospect from a small school.
Conformity
Of course, conformity and social proof are not entirely different in terms of how they function to change our decision-making processes. Conformity has a well documented role in influencing how we behave. While conforming to others in your social group may not function the same as social proof helping you to run from lions, it can change the dynamics of your relationships and influence how you are perceived and treated by others. This is no small thing in a species as socially driven as we are.
If I were to lead into this article with “Bryce Young is a bad football player and should go undrafted” there would be a social cost to this. Colleagues would read it and their opinion of me would be negatively affected. Some people would make disparaging remarks, say I should lose my job or status, or any variety of forms of expressing their view of the reputational damage I’ve done to myself.
Even in situations where there is a small price to pay for avoiding conformity, there is still a large instinctive pressure to do so. Psychologist Solomon Asch’s work on the subject produced one of the most renowned experiments in psychology and has jumped into pop culture to some extent. The experiment, which asked participants to view lines and then select the correct match from multiple choices of varying lengths, shows how strong the social pressure can be to conform even with people you don’t know participating in a low stakes experiment.
Anchoring
Diverging from social psychology into biases and heuristics, anchoring is another area where the consensus or specific pieces of information from others can influence the decision-making of individuals. This functions somewhat differently from social proof or conformity, but instead as a function of how we think about problems when a certain number or option is introduced to our thought process.
If you were preparing to evaluate Zavala only to find that he is ranked as a sixth round prospect by the consensus or that he is 78th on the board of Dane Brugler, you can’t unlearn that when you judge him. No matter what your eyes tell you, there is a large degree of difficulty to pull away from the “anchor” of knowing a player is viewed to be of a certain caliber.
Here is a set of questions for you that were devised by Nobel Prize-winning psychologist Daniel Kahneman to illustrate how anchoring can influence your judgment.
Is the height of the tallest redwood tree more or less than 1,200 feet?
What is your best guess about the height of the tallest redwood tree?
Take a moment to consider this.
The tallest redwood tree is 380 feet tall, but the answers that Kahneman received depending on the first question varied widely. The people given 1,200 feet as their anchoring point guessed around 844 feet on average. Another group given 180 feet as an anchor ended up with an average guess of 282 feet. This is the power of anchoring and it is prevalent in many domains. It is hard to avoid given how our minds process information, but being aware of the role it can have in our judgments can help us to try to work around it systematically.
Here is a video discussing some of the other examples of anchoring studies that have been done, including the sentences given out by judges and real estate value estimates.
Consensus/Markets: Depth, Quality, and Efficiency
In case I’ve been using the word market and consensus too interchangeably, I’d like to define the term a little bit more clearly. When I say, “the market on a player”, I mean the total collection of information on the player as reflected by their ranking on a consensus board. Different consensus boards aggregate in different manners depending on timing, quality of input, and other factors but they all serve relatively similar roles.
This idea of conceptualizing player value through comparison to a market is one that I’ve found to be particularly rich. It allows us to view information in the context of depth, quality, and collective efficiency. Yale economist Robert Shiller’s quotation of one of the earliest definitions of efficient markets is more eloquent than what I could pull off, so I’ll start with that.
There are certainly differences between an aggregation mechanism creating a consensus and a financial market as far as the “number of votes” concept goes, but it’s within reaching distance for us to tie the concepts together. If everyone in a financial market jumps in trying to buy something, the price is likely to rise. If everyone in a community of draftniks and fans begins praising a player and ranking him highly, he will rise on the consensus board.
But there are some issues in these comparisons. Financial markets are often deeper, more informed, and more sensitive to adjustments (due to the skin in the game) than a simpler aggregation mechanism. If you purchase a share of a company in a financial market you are putting your money behind it, whereas anyone could throw together a list of player rankings with no incentive to be accurate beyond the social mechanisms mentioned earlier. You don’t put in five dollars when you call Bryce Young the best quarterback in the draft.
Differences aside though, let’s focus on those aspects of depth, quality, and collective efficiency within a consensus on a player.
Depth
Imagine you and I are sitting on the porch watching birds fly swiftly around the yard. They dive and dart around, landing to pick at the ground in search of worms that have emerged after a recent rainstorm.
I point to the small potted plant sitting against the fence, an area which has not yet been visited by any of these birds. While it may not be as rich in worms as the other areas of the yard, there are surely some worms who are ripe for the taking, assuming a bird is willing to come look.
“How long do you think it will be until the first bird comes over to the potted plant?”
It’s just you and me. I have a guess and you have a guess. Let’s suppose I say 3 minutes and you say 8 minutes. Averaged together, our total guess is that it will take 5.5 minutes for the first bird to arrive.
Now let’s suppose it’s a group of five people, all with relatively similar information. The guesses rattle off. 1. 2. 3. 5. 8.
Our new collective guess is 19 divided by 5 participants, which comes out to 3.8 minutes.
Now let’s suppose the group is twenty people deep, including a couple bird watchers who are particularly savvy to how much time they spend at a particular location. 1. 1. 1. 2. 2. 2. 2. 3. 3. 3. 3. 4. 4. 4. 5. 5. 6. 7. 8. 8.
Our new collective guess is 74 divided by 20 participants, which comes out to 3.7 minutes.
The larger this crowd of guessers gets, the more collective information gets added to the pool and the more useful the consensus is. In most cases the consensus number will be a better predictor than the guess of any individual participant in the crowd. The extremely wrong people on one end tend to offset the extremely wrong people on the other end and the result is an averaging mechanism that provides a good result assuming depth in the market.
In larger markets, there is less volatility when new information is added and there is more value in the information currently present. We know with some degree of certainty that a player who is mocked in the top ten is extremely talented, because we have a large sample of information to draw upon. But if a player only has two or three data points to form their “consensus grade” then we are open to the volatility of a big swing up or down with a new data point. We are also much less certain about their value than we are with the top ten player.
Quality
What about the bird watchers? Aren’t they going to be more accurate than everyone else?
Not necessarily. The aggregation of all total information by the group is a separate entity from any individual guessing. It has features that can’t be captured in the guess of any one person, largely because any one person lacks the information and processing power to effectively and consistently do what a market can do when weighing information.
The bird watcher may guess better than 95% of all members of the crowd, but there will be some crowd members who guess better by dumb luck if the crowd is big enough. That said, the bird watcher’s guess will add significant value to the consensus.
We would likely look in their direction before contributing our own guess for this very reason. If we were asked to place financial wagers on the outcome of a bird visiting the potted plant, we would be cautious in trying to out-predict a bird watcher, or even worse a collection of bird watchers.
Efficiency
I’ll admit, my understanding of market efficiency and the processes with which markets function is not well developed. There is plenty of leg work to do in finding good, thorough reads on the topic. Forgive me for quoting an Investopedia definition in defining market efficiency, but I feel it explains what I’m getting at here very well in terms of the impact of quality and amount (depth) of information generating efficiency.
“As the quality and amount of information increases, the market becomes more efficient reducing opportunities for arbitrage and above market returns.”
If we were to take the group guessing the actions of birds and ask them to place financial stakes on their wager, we would see another shift towards market efficiency as those with low quality information presumably wager cautiously and those with high quality information wager aggressively. This is hard to replicate in a draft consensus setting because the only way we can invoke “aggressive” or “cautious” wagering is by having players higher or lower, not by tying dollar amounts or some other currency of value to our confidence levels.
Why Should I Care?
Here’s the point of all of this talk about markets and consensus. When we consider what the consensus tells us, we have to understand what underlying characteristics it has relative to the prediction it makes on any one player.
I can tell you with near absolute certainty that Bryce Young will be a top pick because the consensus on him has depth, quality, and efficiency. When I look at Chandler Zavala, the consensus prior to Dane Brugler was a couple big boards and mock drafts. It lacks depth, quality (outside of Brugler), and presumably efficiency. This makes trying to understand what we know about him through that consensus extremely volatile and subject to significant changes going forward.
It’s also important to understand this concept when we consider how people are impacted by those previously mentioned aspects like anchoring, conformity, social proof, and private vs. public information. Understanding what comprises public information matters if we are going to use it effectively in making predictions, and we have to be alert for situations where the desire to conform to what others think or follow their lead simply because of our human intuitions is taking us down the path of a possible poor prediction.
Conclusion
Predicting An Outcome (Finally)
Acknowledging all of that uncertainty, I’m going to give my best shot of what happens with Zavala based on the information covered in this article. It would be unfair to take up such a significant amount of your time with theory and jargon without trying to show it actually getting applied in a decision-making process. At heart I’m a practitioner, not a theorist. Put bluntly, none of this shit matters unless we can use it to do a better job of predicting what happens.
On average, 8.33 offensive guards have gone in the first four rounds since 2011. This means that we can use the other guards in the class to help us shape a reasonable expectation, even if we don’t know where Zavala will go. Understanding the full depth of the position group and the number of available substitutes is incredibly important to the exercise of predicting where any one player will go. These guards also have a larger amount of consensus data that has both quality and depth available, so we can lean on this to make projections. Here are the current NFL Mock Draft Database projections.
O’Cyrus Torrence, Florida - 1st Round
Steve Avila, TCU - 2nd Round
Cody Mauch, North Dakota State - 2nd Round
Tyler Steen, Alabama - 3rd Round
Emil Ekiyor, Alabama - 4th Round
Jarrett Patterson, Notre Dame - 4th Round
Andrew Vorhees, USC - 4th Round
Nick Broeker, Mississippi - 5th Round
This matches closely with the average number of guards selected in each round since 2011, so we will pass on adjusting. It is worth noting that players like Peter Skoronski, Darnell Wright, and Matthew Bergeron could be grouped in here, but we’re going to assume they are tackles for this prediction.
Of the players not included on this list, Utah’s Braeden Daniels, Clemson’s Jordan McFadden, UCLA’s Jon Gaines, and LSU’s Anthony Bradford seem like the strongest candidates to break the consensus order. All four tested well, two of them have tackle versatility, and Daniels was a Senior Bowl invitee.
Based on past draft preferences from teams, I’m going to adjust these guys into the mix. Here is my thought process so far.
O’Cyrus Torrence, Florida - 1st Round
No need to change here. Torrence tested questionably and could slide to the second round, but he checks out for everything else outside of that.
Cody Mauch, North Dakota State - 2nd Round
There is a case for Mauch to sneak into the back of the first round given versatility at multiple positions, but I’ll set him in the 2nd. Tested at a high level as expected, had a good pre-draft process overall, and there are no indicators that the consensus will be wrong on him. The versatility and slightly better testing notches him over Avila but it’s 60/40 or 65/35 to me.
Steve Avila, TCU - 2nd Round
No change here. Avila’s testing went as expected and there are no notable indicators to suggest the consensus would be significantly off base.
Tyler Steen, Alabama - 3rd Round
No change here. OT/OG versatility, tested and measured as expected, and no signal for consensus to be wrong here.
Emil Ekiyor, Alabama - 4th Round
This is where it starts to get interesting. Ekiyor chose not to work out by choice and only participated in position drills. He had a very high initial starting point for the consensus because of “too early mock drafts”, so there is a potential for a feedback loop in how he has been ranked. I’m going to keep him here because of the historical averages for guards and lack of suitable alternatives (this would change if shifting over Skoronski, Wright, or Bergeron) but there are reasons to poke at the consensus.
Andrew Vorhees, USC - 5th Round
This list would look much different if Vorhees would have been able to stay healthy throughout the process, but a torn right ACL has impacted his odds of being drafted highly. He tore his ACL on March 5th and the consensus had him at 94th on March 6th (it had not adequately adjusted for the injury). He is now listed at 131st overall as of April 1st. I believe the combination of injury and length will push him slightly below that, so I’m going to say tentatively 5th round with potential to be shifted back up depending on other players.
Nick Broeker, Mississippi - NOT LISTED (Dropping out of 5th)
This is one that sets off some alarm bells for me. Broeker started very high after getting buzz as an offensive tackle last year, and he was included in first round mocks heading into this season. He is somewhat light for a guard at 305 pounds and has short arms and a very small wingspan. I mentioned Patterson’s short arms, and while Broeker’s 32 ½” arms aren’t quite that bad, his 76 ⅝” wingspan is extremely low. Of the 57 guards in last year’s class below a 78” wingspan, none were drafted and only 8 signed undrafted free agent contracts. Broeker will be leapfrogged by some of these other players including Zavala.
Jarrett Patterson, Notre Dame - NOT LISTED (Dropping out of 5th)
Here is another one I’m going to push back against and knock down the list. Patterson had a similar “too early mock” initial ranking that landed him high in consensus, but his testing has been relatively average including a 5.33/5.25 Combine 40 and measuring with extremely poor length. Patterson’s length was a known issue to scouts entering the year, but I don’t think the market has adequately adjusted to it. Having 31 ⅜” arms at guard can be a debilitating issue against certain competition and Patterson’s performance decline in 2022 raises too many red flags for me to keep him here.
Next we have the players who will be moving into this mix.
Braeden Daniels, Utah - 4th Round
Very light at the Combine when he tested well (294 pounds) and was light at the Senior Bowl, but made it to 307 at the Utah pro day. Size is going to be a very big question mark. Top four round testing, OT/OG versatility check out but if teams believe he will consistently hang around the 295 mark as a play weight they may be hesitant. I think he slots in above Ekiyor but below Steen given the versatility and testing. The Senior Bowl invite is worth noting given that the other guys here did not get it, so it can be assumed he was viewed as a top four round type in the fall based on earlier assumptions.
Jordan McFadden, Clemson - 5th Round
McFadden’s height is going to be a very big red flag to some teams but the length, testing, and versatility for OT/OG will be a huge offsetting factor. As of now he is ranked 293th by the consensus though, and there is enough depth to the consensus that I can’t push him any higher than Ekiyor. Vorhees feels like a 50/50 with him post-injury. Shrine Bowl guy as well so lacks a top four round fall element.
Anthony Bradford, LSU - 3rd/4th Round
Bradford’s testing and size are shockingly good and will set him apart from much of the position stack. He is currently ranked 239th, but being an underclassman declaration who ran 5.08/5.02 at 6040 332 while performing admirably in other testing areas sets him up as a potential surprise pick who jumps way out ahead of consensus. I’m going to slot him just ahead of Ekiyor in the 3rd/4th round territory. Personally, in the little bit of what I saw of him I don’t feel comfortable with him going very high, but I’m going to lean on the side of the testing data here.
Jon Gaines, UCLA - 5th Round
Gaines was a late addition to the Shrine Bowl roster, which is pretty telling for how he was viewed entering the pre-draft process, and it’s likely that most teams viewed him as a priority free agent type. He performed incredibly well at the Combine and checks a ton of physical boxes at 6040 303 with 10 ⅛” hands, 33 ⅝” arms, and a 5.01/4.93 40 time to point at among a long list of aced testing categories. For teams that liked him as a mid or late rounder, he may have taken a slight jump upwards, but given the late Shrine invite it’s unlikely that grades from the fall are being completely overwritten. The most likely scenario where Gaines goes incredibly early, at least in my eyes, is if a coach falls in love and fights for him. I’m going to put him in the fifth but slightly behind McFadden and Vorhees.
Here is the updated list now that we’ve gone from purely a consensus ranking to account for information that may not be priced in yet.
O’Cyrus Torrence, Florida - 1st Round
Cody Mauch, North Dakota State - 2nd Round
Steve Avila, TCU - 2nd Round
Tyler Steen, Alabama - 3rd Round
Anthony Bradford, LSU - 3rd/4th Round
Braeden Daniels, Utah - 4th Round
Emil Ekiyor, Alabama - 4th Round
Jordan McFadden, Clemson - 5th Round
Andrew Vorhees, USC - 5th Round
Jon Gaines, UCLA - 5th Round
Nick Broeker, Mississippi - NOT LISTED
Jarrett Patterson, Notre Dame - NOT LISTED
So we’ve got one guy left to add here and this is how we look compared to historical averages right now.
1 - 1st round (1.25)
2 - 2nd round (1.58)
1.5 - 3rd round (2.58)
2.5 - 4th round (2.97)
3 - 5th round (2.25)
Chandler Zavala, North Carolina State - 3rd Round
I haven’t seen official testing data on Zavala, but the stuff I have seen is very kind to him. Even considering that he would be the only non-Combine offensive lineman on this list and a Shrine Bowl player, it’s hard to push back against the information we’ve received from Brugler and the testing data. According to his Shrine Bowl measurements he is 6034 322 with 10 ⅜” hands and 33” arms, all of which matters significantly to a large number of teams. His testing data would suggest that he is one of the best athletes in the class, if it holds up after more reliable sources publish them.
Another case for Zavala’s surprising rise is that he missed a significant amount of 2021 and likely did not receive quite as much attention from teams as players like Avila, Torrence, or Steen/Ekiyor throughout the fall, which could have played a role in not drawing enough love for a Senior Bowl invitation or Combine invitation (although the Combine is a more notable and surprising exclusion).
If I had to give just one singular guess for where he lands, this is where I would expect Zavala to fall in this guard class.
O’Cyrus Torrence, Florida - 1st Round
Cody Mauch, North Dakota State - 2nd Round
Steve Avila, TCU - 2nd Round
Tyler Steen, Alabama - 3rd Round
Chandler Zavala, North Carolina State - 3rd Round
Anthony Bradford, LSU - 3rd/4th Round
Braeden Daniels, Utah - 4th Round
Emil Ekiyor, Alabama - 4th Round
Jordan McFadden, Clemson - 5th Round
Andrew Vorhees, USC - 5th Round
Jon Gaines, UCLA - 5th Round
Nick Broeker, Mississippi - NOT LISTED
Jarrett Patterson, Notre Dame - NOT LISTED
If I were to give a broader guess that had a margin of error, this would be the range, with all bolded players being players that could go ahead or behind him. Vorhees is a wild card given his injury and I can see a world where Avila slightly slides and Zavala jumps him, but I think the bold range is a good approximation of his likely landing spot among the guard class. Let’s call it late second round to late fourth round.
O’Cyrus Torrence, Florida - 1st Round
Cody Mauch, North Dakota State - 2nd Round
Steve Avila, TCU - 2nd Round
Tyler Steen, Alabama - 3rd Round
Chandler Zavala, North Carolina State - 3rd Round
Anthony Bradford, LSU - 3rd/4th Round
Braeden Daniels, Utah - 4th Round
Emil Ekiyor, Alabama - 4th Round
Jordan McFadden, Clemson - 5th Round
Andrew Vorhees, USC - 5th Round
Jon Gaines, UCLA - 5th Round
Nick Broeker, Mississippi - NOT LISTED
Jarrett Patterson, Notre Dame - NOT LISTED
To conclude, I’d say that I believe the current 194th overall ranking is a market mispricing that will be closed a lot more towards the 60-130 range as we get closer to draft night. While there is certainly the possibility of an information cascade going on, the testing data and physical profile is a great match for an early round lineman based on historical norms and should, at least to some degree, offset the low initial signals of the Senior Bowl and Combine passing him up. Those low initial signals are likely what prevented him from having an earlier rise for the consensus, as most rankings that included him prior to March 21st were lower in both quality and quantity. Brugler’s incentives and private information, as reflected by what he does publicly, would suggest that his ranking is a valuable one as well and we should consider that as a positive for Zavala.
Making Sense Zavala This
Part of what stoked my interest so much with Zavala is that he was not a player who was on my radar or someone that I’ve watched in any detail. I won’t lie, I’ve taken a peek since Brugler tweeted about him, but I don’t have any wholly formed thoughts about him as a player based on the tape. He sits at a beautiful intersection of time, information, expert impact, and peculiar signals throughout the process. He is, as much as it can be expressed, exactly what makes the draft process so fascinating to watch unfold.
While I don’t believe this article will change the minds of anyone who watched Zavala and either liked or disliked his game, I hope it can provide an interesting look at how we think about both him and every prospect when put under the microscope of the information known about them. If the goal is to make good predictions, we must look through a wide lens and try to understand problems for how truly complex they are, not just what comes easily to our eyes and mind when we watch a player, look at his stats, or see those in our social tribes projecting certain things for him.
The draft, both on an individual player level and on the level of every system within it, is a vast web of public and private information, path dependent occurrences, feedback loops, social interchanges, and more all wrapped into one unfathomably large package of uncertainty.
To treat it as anything less than that would be doing a disservice to the challenge we face in trying to make the highest quality predictions that we can about it.
Resources
Update 4/7/23
Official numbers have been released on Zavala. The 5.01 40 turned out to be incorrect and his overall testing was slightly less impressive, although his overall profile still turned out to be pretty good. I'm going to stick with the original prediction as far as where he lands on draft night being somewhere between 60-130, although I think there's more room for a slight dip on the back end with those numbers.
Brugler's private information (as reflected by his ranking) still outweighs slightly worse testing than originally expected. Posted below is the updated RAS on Zavala according to the numbers released on draftscout.com this morning. If the original prediction turns out to be significantly off base I'll post a follow up and try to figure out what I missed.
Comments